受限领域智能问答系统的研究与实践

智能问答技术是自然语言处理领域中一个非常热门的研究方向,它综合运用了各种自然语言处理技术.本文主要研究智能问答系统的关键技术.首先,详细地介绍了自动问答系统的概念原理与主要组成部分.其次,对该系统的中文分词模块及其相关技术进行了比较深入的研究,选择最大匹配、逆向最大匹配和设立自然切分点相结合的算法;在歧义字段处理方面,提出了基于句子整体考虑的计算句子词频和的方法,能够在处理特定歧义问题方面有很好的改进和提高.再次,对特定领域的问题进行分类,便于系统根据问题类别进行答案匹配.最后.利用赋权值方法,完成答案的提取,以句子作为答案匹配的基本单位.基于上述理论与算法,实现了一个地理领域的智能问答系统.

Study on Intelligent Question-answering System of Restricted Field

Question-answering technology is a hot research field in natural language processing. QA includes many kinds of NLP technologies. This paper mainly researches the pivotal technology of Q/A system. Firstly, it introduces the conceptual theory and parts of intelligent question-answering system in detail. Secondly, It researches the Chinese word segmentation module of this system and its relevant technology, including the maximum match, reverse maximum match algorithm with natural syncopate point. In the aspect of different meanings, it puts forward the method of calculating sum of word frequency based on the whole sentence. It makes a better improvement for processing the problems involving different meanings. Thirdly, it brings forward the classification problem of specific fields, which is convenient for result matching according to the kind of problem of the system. In the end, this system completes the results extraction by calculating the weight. It makes the sentences be the basic unit of the result matching. On the basis of theory and algorithm mentioned above, a question-answering system in geographic field is implemented. The intelligence degree of question-answering is considerably enhanced.